Preclinical studies of radiopharmaceuticals for positron emission tomography (PET) are critical for their translation from the radiochemistry research lab to their use to perform physiologic measurements in humans. Areas of preclinical research include performing radiation dosimetry calculations to determine the radiation exposure subjects will receive, and the development and validation of tracer kinetic models that describe the in vivo behavior of the compounds. The PET Department has developed [Tc-94m]sestamibi to image multidrug resistance in human cancer. Complex radiation dosimetry calculations were performed in preparation for these studies. Tc-94m is produced by a CS-30 cyclotron and incorporated into [Tc-94m]sestamibi. Eight other Tc radionuclides with varying half-lives are also produced, and also label sestamibi. The radiation exposure from sestamibi labelled with all of these radionuclides was calculated. Because five of the radionuclides are not in the standard MIRDOSE software used to calculate radiation exposure, special manual computation procedures were required. The organs with the largest exposures were found to be small intestine (1.11 rad/mCi), and upper large intestine and urinary bladder (1.01 rad/mCi). The dosimetry was incorporated into a National Cancer Institute protocol to study multidrug resistance in cancer patients. [C-11]Flumazenil is a benzodiazepine receptor antagonist widely used to image brain benzodiazepine receptors with PET. Existing radiation dosimetry calculations, based on data obtained in mice, were deemed inadequate for clinical research projects. In collaboration with investigators from NIMH, whole body PET studies were performed in humans to determine the biodistribution of [C-11]flumazenil and to obtain radiation dose estimates. The critical organ was found to be urinary bladder (radiation exposure of 0.136 rad/mCi), followed by GB wall (0.0338 rad/mCi), brain (0.0333 rad/mCi), and liver (0.0311 rad/mCi). For these organs, the radiation exposure based on human biodistribution data exceeded the published results based on mouse data by factors of 1.3 to 3.0. These results demonstrated the limitations of using rodent biodistribution data.